Cumulative Overdue Model of Employees’ Housing Accumulation Fund Loans Based on Decision Tree

Abstract

To predict and control the risks that the housing accumulation fund loans taken out by employees will be cumulatively overdue, we selected the cumulative overdue rate as the target variable and 31 independent variables based on the data on housing accumulation fund loans of City T over the years. We did some data-mining using the decision tree and found 62 rules that apply to a “low risk†situation and 12 rules that apply to a “high risk†situation. The classification accuracy of the rule set on the testing data set is 81.72%. Therefore when considering an loan application, we can use this rule set to predict the risk that the loan will be overdue and be better informed in credit evaluating.